Sensitivity-n evaluation source code

marcoancona / DeepExplain

A unified framework of perturbation and gradient-based attribution methods for Deep Neural Networks interpretability. DeepExplain also includes support for Shapley Values sampling. (ICLR 2018)

https://arxiv.org/abs/1711.06104

MIT License

720 stars 133 forks source link

Sensitivity-n evaluation source code #69

Open Subh1m opened 2 years ago

Subh1m commented 2 years ago

@marcoancona @Enny1991 @harkous @yuhui-zh15

Can you please show which part in the code refers to the sensitivity-n evaluation?

Subh1m commented 2 years ago

I found the code in the shapely branch. It seems we are creating a deltas.hdf5 file for the sensitivity-n test. Can you please explain how to understand the results? I extracted the data from the hdf5 file and got an array for each attribution method.